Flooding poses a significant challenge to urban development, exacerbated by climate change and increased urbanization. This study explores the potential of artificial intelligence (AI) to improve flood severity forecasting, thereby aiding sustainable urban planning. We conducted a comprehensive comparative analysis of various AI models—including traditional regression techniques, ensemble methods, and advanced deep learning architectures—to enhance flood prediction accuracy and support resilient urban design. Using datasets from 25 diverse states encompassing a wide range of urban conditions and flood-related variables, we evaluated the performance of models such as Decision Trees, Random Forests, Gradient Boosting Machines, Support Vector Regression (SVR), Lasso Regression, time-series models like ARIMA and SARIMA, and neural network architectures including Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Long Short-Term Memory networks (LSTM). The results demonstrate that deep learning models significantly outperform traditional approaches in predicting flood severity, evidenced by the lowest mean squared error (MSE). This suggests that AI-driven models can be effectively integrated into urban planning frameworks to enhance flood management strategies and promote sustainable development. This investigation highlights the transformative role of AI in advancing urban planning through improved flood forecasting, providing critical insights for urban planners, policymakers, and researchers, and emphasizing the importance of employing advanced AI techniques to tackle environmental challenges and build resilient cities.